F@H on Linux problem

[linkinpark342]

Hey,
I'm having a problem running folding on my gentoo computer. i have two cores running and one of them seems to work fine but the other one just keeps failing with an error message:

[12:43:40] Completed 50000 out of 250000 steps (20%)
[12:43:40] - Checksums don't match (work/wudata_08.xtc)
[12:43:41] Premature end of file when checksumming (170264 bytes left)
[12:43:41] - Could not calculate checksum (work/wudata_08.xtc)
[12:43:42] Checksum not what expected.
[12:43:43]
[12:43:43] Folding@home Core Shutdown: FILE_IO_ERROR
[12:43:45] CoreStatus = 75 (117)
[12:43:45] Error opening or reading from a file.
[12:43:45] Deleting current work unit & continuing...

I just don't understand why one of my computers fails while the other one just keeps on chugging. thanks in advance

 

[CupCak3]

what FS are you using and also you should probably run a chckdsk on your hdd

 

[GLeeM]

I shouldn't even but in here - I don't know Linux - but is there enough disk space or enough disk space assigned to the program or directory?

 

[Coolkid]

Make sure you are running them in different directories, and also check that the user running F@H has read/write access to the folding dir and files.

 

[linkinpark342]

I'm using ext3 for where it should be saving to and have about 13G available so that shouldn't be a problem.
[edit: p.s.
client1(i.e. core 1) says that error message while client2 just gave me a nice [23:23:58] Completed 3750000 out of 3750000 steps (100%)]

 

[GLeeM]

Is this on one computer or two computers?

If on one computer - when running two instances, you need to configure the second instance to a different MachineID.

The client defaults to MachineID #1.

Stop the client and restart with the "-config" flag/switch. The last question asked (at least with the Windows version) is where you can change the MachineID number, just change it to "2".

 

[linkinpark342]

I believe when you run the /usr/bin/foldingathome file (or the /etc/init.d/foldingathome like i do) it runs a script that automatically copies the cfg to each directory and increments the machine id (this is an assumption that i am making on the evidence that, well, the cfg in the second thread IS incremented but i'm not sure if its using that cfg or the "client2.cfg" file in the same folder)

 

[GLeeM]

OK, I am going to give a bump here so that some Linux folders will see and give you a hand.

It looks like your troubles could be because of two clients trying to write to the same files, in other words, they have the same MachineID.

Please be patient, help will be along.

And welcome to the TeAm

 

[biodoc]

Originally posted by: linkinpark342
I believe when you run the /usr/bin/foldingathome file (or the /etc/init.d/foldingathome like i do) it runs a script that automatically copies the cfg to each directory and increments the machine id (this is an assumption that i am making on the evidence that, well, the cfg in the second thread IS incremented but i'm not sure if its using that cfg or the "client2.cfg" file in the same folder)

Here's how I do it (very similar to the windows client set up):

-remove your current installation!!
-to run 2 clients you need a cpu that is HT capable or a dual core cpu, etc.
-I run fah from my home directory (no need to clutter up /usr/bin, etc with these kinds of executables!)
-from a terminal in your home directory, create 2 directories, for example fh1 & fh2
- download a copy of the linux client to each directory (fh1 & fh2)
-from a terminal, change directory to fh1
-make sure the fah client is executable: ./chmod 700 FAH502-Linux.exe
-run the client and configure: ./FAH502-Linux.exe -config
-respond to the questions (make sure the team is 198!); IMPORTANT: answer yes to "change advanced options"
-respond to "machine ID": "1" which is default
-finish responding to the questions then the client should run as normal

-create another terminal window and change directory to fh2
-follow the above instructions with one exception!! respond to "machine ID": 2

another point: apparently the -local switch is meaningless in linux

Hope this helps!!

 

[linkinpark342]

Thanks for the bump GLeeM and for the tutorial biodoc. I eventually got it to work by re-emerging it. and changing the /etc/conf.d/foldingathome file or whatever it is from autodetecting number of processors to hardsetting it to 2. I <3 emerge . I had it working with your tutorial but i wanted it to run as a service sorta for the linux box so instead of learning all about the runlevel stuff i thought i'd give the emerge another chance and it works . Thanks again.

 

[biodoc]

Glad you're up and running!